Search

Collision-free control and planning for autonomous vehicles in an only partially known environment is challenging. Often this problem is tackled passively - without active exploration, i.e., planning and control are performed based on the information obtained during operation without explicitly considering that reducing the environment uncertainty by active exploration can improve the overall performance, e.g., reduce the time needed to achieve the goal. We propose a moving horizon planning approach combined with a dual-mode predictive control formulation that actively explores the environment to obtain improved obstacle information. Given a path by the moving horizon planner, the dual-mode predictive tracking controller balances two objectives: fast and safe goal achievement, taking into account potential improvements due to active exploration - learning - of the environment. Adding an excitation term to the objective function incentivizes the system to explore the environment to enhance the overall objective and reduce uncertainty. The exploration-exploitation formulation enhances the overall performance while ensuring obstacle avoidance in combination with the moving horizon planning strategy. An example of an autonomous ground vehicle operating in a cluttered, only partially known environment demonstrates the approach's efficacy.

Model-based control of biotechnological processes is, in general, challenging. Often the processes are complex, nonlinear, and uncertain. Hence modeling tends to be complex and is often inaccurate. For this reason, non-model-based control strategies developed via fask, bench-scale, or pilot plant experiments are often applied in the biotechnology industry. Model-based control and optimization techniques can increase processes’ performance and automation level, thereby decreasing costs and guaranteeing the desired specifications. These rely on a model of the process to make predictions and optimize the inputs to the plant. To improve the quality of the models, it is often helpful to use combined first principle and data-driven models together in a hybrid modeling approach which increases the model prediction capabilities. The residual uncertainty of the hybrid model should be taken into account in the control level to satisfy the process specifications and constraints. This paper proposes to use a stochastic model predictive control scheme that exploits a hybrid, Gaussian processes-based model. We outline the effectiveness of the stochastic model-based approach in combination with a suitable Kalman filter for state estimation considering an example biotechnological process. Furthermore, we underline that appropriate tools exist that allow the simple application of such methods even for the novice user. To do so, we use an open-source Python package — HILO-MPC, which allows the simple yet efficient formulation and solution of machine learning-supported optimal control and estimation problems.

Competitive biotechnological processes need to operate over various conditions and adapt to changing economic contexts. Dynamic ATP turnover allows trading off declines in biomass formation and volumetric productivity for enhancements of product yields in fermentations where the product pathway is linked to ATP synthesis. To facilitate its practical implementation, we propose to dynamically manipulate the cellular ATP turnover by putting the ATPase enzyme, which hydrolyzes ATP into ADP, under the control of an optogenetic gene expression system. This allows achieving dynamic control of the ATP wasting online via a tunable external input. While light as control input is promising because it is easily tunable, generally non-invasive/non-toxic, and more affordable than classical chemical inducers, it makes the overall control task challenging. Thus, we derive an expanded version of dynamic enzyme-cost flux balance analysis that takes into account the dynamics of the optogenetic actuator. We then formulate a suitable optimal control problem to find optimal inputs for achieving the desired process performance. We test our approach in simulations using the batch anaerobic lactate fermentation of glucose by Escherichia colias a case study.

Precise and accurate online monitoring methods are needed to enable smart biomanufacturing and automation. Most of the sensors available focus on process parameters such as metabolite and dissolved gas concentrations, cell density or viability, among other variables like pH, temperature, etc. In this work, we develop a soft sensor algorithm to estimate the cell composition online, a very important aspect often overlooked in the bioprocess monitoring literature. Our strategy is based on full information estimation, an optimization-based estimator that takes into account the dynamics of the cell metabolism and considers all the available measurements from the beginning of the process, thus it has a memoryeffect. Being able to track dynamic changes in cell composition can open the door to promising applications, e.g., predictive control and automation of biosystems. As a case study, we consider the Escherichia coli’smetabolism growing on glycerol under different levels of oxygen supply. We compare the performance of our soft sensor method against resource balance analysis, a previously proposed estimator based on steady-state assumptions. Overall, the presented full information estimator was able to track the dynamic changes in cell composition significantly more accurately. We also discuss how our estimation strategy can be transformed into a moving horizon estimation, where only the available measurements in a fixed and moving window are considered, thereby reducing possible computational burdens.

Cell-free biosynthesis uses the machinery of cells, such as the metabolic reactions, to carry out conversion processes in vitro. This can be more beneficial than in vivo approaches like fermentations. Some advantages of these synthetic biology processes include higher product yields, rates and titers, and more flexibility in pathway design. Cell-free biosynthesis is still in early stages and, unlike in vivo production, there are very few examples of model-based optimization. Moreover, we encounter static optimizations in most cases, neglecting the dynamic nature of the processes. We present an optimal control framework to maximize the efficiency of cell-free biosynthesis. We focus on fed-batch setups as they allow enhancing the reaction rates via the feeding, extending production processes for longer times, and minimizing the potential negative effects of enzyme kinetics with substrate inhibition. Our framework can in principle handle several cost functions and exploit both static and dynamic degrees of freedom. An aspect that can hinder model-based optimization is model uncertainty, which can arise due to uncertain parameters, oversimplified model assumptions or unknown reaction mechanisms. To counteract this, we propose the use of model predictive control during the process operation. In addition, we outline the use of moving horizon estimation as an observer in the case of unmeasured states. We consider the de novo cell-free synthesis of uridine diphosphate-N-acetylglucosamine as a biomedical relevant case study, where we were able to maximize the volumetric productivity in simulations, and indirectly also the titer and enzyme efficiency use.

Many processes operate repetitively, for example batch processes in biotechnology or chemical engineering. We propose a method for risk-aware run-to-run optimization and model predictive control of repetitive processes with uncertain models. The goal is to increase the performance as the number of runs increases by improving the model despite limited measurements while considering model uncertainty and avoiding uncertain areas. The method uses a gray-box model, i.e. a model formed by a first principle and a machine learning component, in this case an artificial neural network. The model uncertainty might be large, particularly in the first runs, where only a few measurements are available. We propose to quantify this uncertainty using Bayesian inference. This is in turn reflected by a risk measure entering an open-loop optimal control problem and a shrinking-horizon Model Predictive Controller as a constraint to limit control and exploitation in high risk areas. We show that using this risk measure we are able to efficiently reach high process performance. The proposed method is tested in simulations on two biotechnological fed-batch processes.

Safe autonomous passing of intersections with mixed traffic, including human drivers and autonomous vehicles, is challenging. We propose a tailored approach that provides guarantees despite uncertainties fusing learned models and model predictive control. A single autonomous vehicle is controlled by the predictive controller via acceleration and steering angle without assumption of a global controller. Each maneuver of the human behaviour is modeled with a neural network, which enters the predictive controller formulation as a constraint. As an example, we consider a single autonomous vehicle on an unsignalized intersection, which gives right-of-way to a human-driven vehicle. We show how human driving behavior can be modeled based on real recorded trajectory data and implemented in the proposed predictive control approach by dynamically changing the constraints of the optimization problem.

Dynamic models of biotechnological processes form the basis of process optimization, control, and estimation. Metabolic network models are often at the core of such models. Since metabolic network models can be very large, and consequently computationally expensive, model reduction techniques can be applied. The derivation of a suitable reduced metabolic network that captures the essential metabolism is still a challenging problem. State-of-the-art network reduction algorithms utilize a priori defined phenotypes that reflect the expected behavior of the biological system. However, most bioprocesses undergo changes in the metabolism, hence, a switch in the cellular phenotype. If these phenotypes are unknown a priori, the reduced network fails to represent all observed metabolic behaviors. Contrary to these approaches, we propose a method that reduces genome-scale metabolic networks models using data from real experiments instead of relying on predefined phenotypes. Doing so, we circumvent the use of a priori information and guarantee that the network is capable to describe all observed phenotypes and can be reliably used for estimation, prediction, and optimization.

In upstream Oil and Gas operations a well is drilled following a planned trajectory. The trajectory is designed to avoid hard formations and other wells while minimizing drilling time. The uncertainty of the environment, e.g. unknown rock hardness, effects negatively the efficiency of operation: drilling time increases due to frequent corrective control actions that must be taken to counteract disturbances and risk increases since process constraints may be violated. This paper proposes an event-triggered multi-stage model predictive control that aims at tackling both challenges. The event-triggering strategy tries to minimize the number of control actions sent to the actuators, while the multi-stage strategy improves constraints satisfaction despite uncertainties. The method is tested in simulation where unknown changes in rock hardness are considered. In comparison to a standard model predictive control approach, we show that using the combined event-triggered and multi-stage approach we improve constraints satisfaction and decrease the number of control actions.

Robust control of uncertain nonlinear systems subject to constrants often leads to conservatism. Such behaviors can be improved by updating the model of the uncertainty with the data collected during the operation time or by bounding the parameters. This paper proposes an approach to robustly control the discrete-time Lur’e system subject to states and input constraints, where the unknown memoryless nonlinearity is sector-bounded and its Lipschitz constant is assumed to be given. In the first phase of operation, when no data has been collected, a robust MPC controller obtained from solving linear matrix inequalities is used. This formulation is also used to compute the safe region, in which the system can operate safely. After sufficient data has been collected, an upper and lower bound of of the nonlinearity can be constructed by using the Lipschitz constant. A controller based on tube-based MPC is used, which results in less conservatism and provides more flexibility. Data of the nonlinearity can be further updated to reduce uncertainty, and hence, to decrease the size of the tube. Under additional conditions, the controller can safely explore the region outside the safe regions to collect more data of the unknown nonlinearity to improve performance and region of attraction.

Biotechnological processes are urged to become more flexible and sustainable in order to meet the challenges of modern society. Using renewable feed stocks for the production of value added products is one strategy to achieve sustainability. The challenges arising from the use of renewables are large compared to the commonly used resources. The concentration of nutrients in renewable feed stocks varies between batches and is not optimal in respect to the demands of the cells. The biological system adapts to this situation by changing the metabolic growth modes in dependence of the availability of nutrients in the media. Consequently, the process can run through multiple modes. Each switch of the mode results in a change of the system dynamics which effects the process performance. To overcome the challenges, we propose a model predictive control approach combined with a moving horizon estimator that takes directly the multi-mode nature of the process into account. It ensures optimal performance while guaranteeing that the constraints are met in each phase of the process. The approach is motivated by and applied to a sustainable biopolymer production from juice waste.

Many systems exhibit multiple modes of operation, where the real mode is not known a priori. Examples are the grasping of objects by a robot, where a variety of object forms and stiffnesses are possible, or a quadcopter lifting and dropping several objects with different weights. In such cases it is challenging to design a controller that ensures robustness and safety for all possible modes without jeopardising performance. To this end, we present a learning supported predictive control approach for multi-mode uncertain environments with robustness guarantees. The approach allows to fuse a priori knowledge via including multiple models of the system with on-line and off-line learning to improve the system models and thus the overall performance. Guaranteed robustness is ensured by decoupling it from performance. While learned and improved models are used for performance optimization, robustness is guaranteed by ensuring that all possible modes respect the constraints and are repeatedly feasible. The ideas are confirmed considering an UAV package delivery example.

The most recent information on fascial tissue indicates that there are not fascial layers, but polyhedral microvacuoles of connective tissue, which connect the body systems and, by hosting specialized cells, permit several functions, such as motor, nervous, vascular and visceral. These microvacuoles (a repetition of polyhedral units of connective fibrils) under internal or external tension change shape and can manage the movement variations, regulating different body functions and ensuring the maintenance of efficiency of the body systems. Their plasticity is based on perfect functional chaos: it is not possible to determine the motion vectors of the different fibrils, which differ in behavior and orientation; this strategy confers to the fascial continuum the maximum level of adaptability in response to the changing internal and external conditions of the cell. The present commentary deals with this concept, providing clinical examples of different disease patterns, providing contrary examples in which this adaptability does not occur, and lastly suggesting considerations for the approach to manipulative therapy of the fascial tissue. The fascial continuum is like a flock of birds flying together without a predetermined logic and maintaining their individuality at the same time.

Patients benefit from the presence of empathic caregivers (CGs). In this regard, empathy toward the patient is one of the clinical targets for improving patient outcomes. However, relatively little is known about the impact of patients' empathic responses on the CGs' burden. Among people living with Parkinson's disease (PwP), care partners play a major role. This study involved 28 spouse-patient couples. Empathy, stress burden, and mood disorders (such as anxiety and depression) were assessed over a 6-month period, before and after the reported intervention. Our observation points out that the improvement of patient empathy is necessary for a significant burden reduction among spouses caring for PwP.

By now many results with respect to the fast and efficient implementation of model predictive control exist. However, for moving horizon estimation, only a few results are available. We present a simple solution algorithmtailored to moving horizon estimation of linear, discrete-time systems. In a first step the problem is reformulated such that only the states remain as optimization variables, i.e. process and measurement noise are eliminated from the optimization problem. This reformulation enables the use of the fast gradient method, which has recently received a lot of attention for the solution of model predictive control problems. In contrast to the model predictive control case, the Hessian matrix is timevarying in moving horizon estimation, due to the time-varying nature of the arrival cost. Therefore, weoutline a tailored method to compute online the lower and upper eigenvalues of the Hessian matrix required by the here considered fast gradient method. In addition, we discuss stopping criteria and various implementation details. An example illustrates the efficiency of the proposed algorithm.

The definition of fascia includes tissues of mesodermal derivation, considered as specialized connective tissue: blood and lymph. As water shapes rocks, bodily fluids modify shapes and functions of bodily structures. Bodily fluids are silent witnesses of the mechanotransductive information, allowing adaptation and life, transporting biochemical and hormonal signals. While the solid fascial tissue divides, supports, and connects the different parts of the body system, the liquid fascial tissue feeds and transports messages for the solid fascia. The focus of this article is to reconsider the model of biotensegrity because it does not take into account the liquid fascia, and to try to integrate the fascial continuum with the lymph and the blood in a new model. The name given to this new model is RAIN—Rapid Adaptability of Internal Network.

Every body structure is wrapped in connective tissue or fascia, creating a structural continuity that gives form and function to every tissue and organ. The fascial tissue is uniformly distributed throughout the body, enveloping, interacting with and permeating blood vessels, nerves, viscera, meninges, bones and muscles, creating various layers at different depths and forming a tridimensional metabolic and mechanical matrix. This article reviews the literature on the emission of biophotons and adjustable sounds by the fascial system, because these biological changes could be a means of local and systemic cellular communication and become another assessment tool for manual (therapy) practitioners. This is the first article that discusses these topics in a single text, attempting to bring such information into an area of application that is beneficial to osteopaths, chiropractors, and manual therapists.